Combined CD+PMD might be interpreted in an ambiguous way

[jktjkt]

When the model specifies a combined CD and PMD penalty, some data might be ambiguous. For example, if the transponder performance model provides a combined penalty for the 🔴 red dots only, how do we interpret the value at the 🟢 green point?

        0 ps/nm       4000 ps/nm                     12000 ps/nm
           |             |                                |
           v             v                                v

 0 ps -->  🔴------------🔴--------------------------------🔴-------------> CD
           |
           |
10 ps -->  🔴
           |
           |
           |                                   🟢
           |
           |
30 ps -->  🔴
           |
           |
           v
          PMD

In case where only the data on axes are given, then the intention was probably to infer the 🔵 blue points as a simple linear combination of these two penalties. In other words, in that case we can treat both penalties as independent ones. It would be nice to have an explicit way of defining that, and the suggestion made on call to use an union with the empty data type and rely on empty for the "other" penalty instead of using a regular 0 value will probably address this problem. Anyway, we can probably use some linear approximation in between the provided + inferred values:

        0 ps/nm       4000 ps/nm                     12000 ps/nm
           |             |                                |
           v             v                                v

 0 ps -->  🔴------------🔴--------------------------------🔴-------------> CD
           |
           |
10 ps -->  🔴            🔵                                🔵
           |
           |
           |                                   🟢
           |
           |
30 ps -->  🔴            🔵                                🔵
           |
           |
           v
          PMD

However, a more interesting example happens when the input data are provided for both (CD, PMD), and when the input does not explicitly cover its full "bounding rectangle", like the next example. What is the intention here -- does it mean that the transponder only supports transmission when the combined CD/PMD is "inside the area" that's covered by the red dots? Or should we still somehow combine the most extreme red dots as before? Do we have a guarantee that a penalty for "highest allowed CD, no PMD" and a penalty for "highest allowed PMD, no CD" are both always present, and that we can combine these two values together and interpolate as before?

           🔴------------🔴------------------------------------🔴---------> CD
           |
           |
           🔴            🔴                        🔴
           |
           |
           |                                   🟢
           |
           |             🔴
           |
           |
           🔴
           |
           |
           v
          PMD

I'm filing this issue in case ietf-ccamp-wg/ietf-ccamp-layer0-types-ext#50 results in keeping the combined CD+PMD around. Perhaps a lot can be achieved by some careful wording in the draft (or perhaps even in the YANG descriptions) to reduce this ambiguity?

[sergiobelotti]

02/22/2022 Call: We will change YANG model to have a list of CD/penalties and PMD/penalties separate list , if there is not positive feedback regarding possible publication justifying the triplet solution.

[sergiobelotti]

Could we consider close the issue and take only the above one related to layer0-types-ext https://github.com/ietf-ccamp-wg/ietf-ccamp-layer0-types-ext/issues/44 ?

We can consider closed the issue as soon as we will have the PR for IETF-114